Uptime and heartbeats
Know whether the release held with availability checks.
Auto-detectTimelinePostmortemMaintenance
Detect, coordinate, and close every incident.
When a monitor fails, the incident is already open and deduped. Run the response on a live timeline with a lead and responders, then close it out with a governed postmortem and a scheduled maintenance window.
Checkout API degradation
p1major#231
leadA. RiveraresponderJ. Okafor
- monitoringpublicjust now
Error rate back to baseline. Watching before we resolve.
— A. Rivera
- identifiedinternal6m ago
Traced to a saturated connection pool on the checkout service.
— J. Okafor
- investigatingpublic14m ago
We are investigating elevated errors on checkout.
— A. Rivera
Incidents turn a detected outage into a coordinated response.
One connected loop, held on the stage this capability serves. The other stages stay as context so you can see what feeds in and what comes next.
BD
Discovery
Feedback
Product
Engineering
Release
Reliability
Evidence
Reliability
Know whether it held.
Available now
Incidents and status pages
Keep incident response and customer-facing status context together.
Web performance
Review Lighthouse report evidence against configured targets.
Test intelligence
Investigate test outcomes and flakiness evidence across runs.
Automatic detection, deduped into one incident.
A failing uptime monitor opens an incident on its own. Repeated failures from the same monitor fold into that single incident instead of paging you per check, and when the monitor recovers it auto-resolves.
checkout-api
failing- check failed · 50214:02:11
- check failed · 50214:03:11
- check failed · timeout14:04:11
deduped into one
Checkout API degradationopened
#231 · automated · one incident, not one page per check
on recovery it auto-resolves
Every incident carries its blast radius.
The incident knows which monitors it affects and their current status, which public status pages it appears on, and how many people subscribed for updates. Scope it to the whole organization or to a specific list of monitors — this is impact accounting, not a guess at cause.
#231 impact
scoped monitorsaffected monitors
- checkout-apidown
- payments-apidegraded
- web-frontendup
shows on status pages
- Public status/status
- Partner status/partners
subscribers on this incident12
One live timeline, the rest is history.
Only the incident still in flight moves.
Every incident sits on a time-ordered rail carrying its severity and priority. Resolved ones go quiet; only the active one pulses, so the thing that needs you is the thing you see.
p1 to p4critical to cosmeticreopen chain
incident history
- Elevated 5xx on api#218 · p2 highresolved9d ago
- Certificate renewal#224 · p4 lowresolved5d ago
- Checkout API degradation#231 · p1 criticalactivenow
Open, work it, resolve — and reopen if it comes back.
An incident moves through a clear set of states. A resolved incident is never truly gone: if the same problem returns, it reopens as a linked follow-up so the history stays connected.
automated or manualresolvereopen as linked
incident state
- 1open
- 2acknowledged
- 3resolved
resolved → reopened as a linked follow-up
The write-up and the next window are here too.
A postmortem you have to approve before it ships.
Write the postmortem in markdown and move it through draft, approved, published, and withdrawn. Publishing needs approval first, and every save is versioned so two editors cannot quietly overwrite each other.
draft → approvedpublishedwithdrawn
Checkout API degradation — postmortem
rev 4- draft
- approved
- publishedhere
- withdrawn
approval before publish · versioned saves
Schedule the work, mute the noise, tell the world.
A maintenance window covers a fixed span and the monitors it touches. Switch on alert suppression so planned work does not page anyone, and show it on your status page so customers know it is planned.
scheduledsuppress alertson status page
DB migration
scheduledSat 02:00Sat 03:30
- Suppress alerts
- Show on status page
affected monitors
checkout-apidb:5432 (tcp)nightly-sync
What this is
Incidents is the coordination record: detection, a timeline, responders, a governed postmortem, and maintenance windows. It does not guess at root cause or wire your outage to a deploy — it keeps repeated failures from the same monitor as one honest incident and gives your team one place to run it.
The vocabulary of an incident.
The states, severities, and controls a real incident moves through — the same set the dashboard uses.
Severity & priority
Two independent scales.
criticalmajorminorcosmeticp1p2p3p4
Update timeline
Each update, public or internal.
investigatingidentifiedmonitoringresolvedpublicinternal
Response & scope
Who and what it covers.
leadrespondersubscribersorganizationspecific monitors
Postmortem
Governed and versioned.
draftapprovedpublishedwithdrawn
Maintenance
Planned, not surprising.
scheduledactivecompletedcancelledsuppress alerts
Source
How it started.
automatedmanualdedupedauto-resolvereopen
What feeds incidents, and what carries them out.
Be ready before the next one.
Start free
Connect a monitor and your incidents open themselves.